BCL-6, a POZ/Zinc-Finger Protein, Is a Sequence-Specific Transcriptional Repressor CHIH-CHAO CHANG*, BIHUI H

Proc. Natl. Acad. Sci. USA Vol. 93, pp. 6947-6952, July 1996 Biochemistry

BCL-6, a POZ/zinc-finger protein, is a sequence-specific transcriptional repressor CHIH-CHAO CHANG*, BIHUI H. YE*, R. S. K. CHAGANTIt, AND RICCARDO DALLA-FAVERA*t *Division of Oncology, Department of Pathology, and Department of Genetics and Development, College of Physicians and Surgeons, Columbia University, New York, NY 10032; and tCell Biology and Genetics Program, Memorial Sloan-Kettering Cancer Center, New York, NY 10021 Communicated by Robert C. Gallo, University of Maryland, Baltimore, MD, March 22, 1996 (received for review December 29, 1995)

ABSTRACT Approximately 40% of diffuse large cell lym- mental regulators Tramtrack and Broad-complex (20, 21), the phoma are associated with chromosomal translocations that human KUP (22), ZID (19), and PLZF (23) proteins as well as deregulate the expression of the BCL6 gene by juxtaposing by POX viruses (24) and the actin-binding Drosophila oocyte heterologous promoters to the BCL-6 coding domain. The protein Kelch (25). These structural features and the expression BCL6 gene encodes a 95-kDa protein containing six C- pattern of the BCL-6 protein suggest that it may function as a terminal zinc-finger motifs and an N-terminal POZ domain, transcription factor involved in the control of B-cell development. suggesting that it may function as a transcription factor. By This study was aimed at elucidating the role of BCL6 in using a DNA sequence selected for its ability to bind recom- transcriptional regulation. We demonstrate that BCL6 func- binant BCL-6 in vitro, we show here that BCL-6 is present in tions as a potent transcriptional repressor of promoters linked DNA-binding complexes in nuclear extracts from various to its DNA target sequence and we map this transrepression B-cell lines. In transient transfection experiments, BCL6 can activity to two noncontiguous N-terminal regions of the pro- repress transcription from promoters linked to its DNA target tein, one of them including the POZ domain. sequence and this activity is dependent upon specific DNA- binding and the presence of an intact N-terminal half of the MATERIALS AND METHODS protein. We demonstrate that this part of the BCL6 molecule Plasmids. The eukaryotic expression vector pMT2-BCL-6 has contains an autonomous transrepressor domain and that two been described (12). To construct the plasmids expressing the noncontiguous regions, including the POZ motif, mediate Gal4-BCL-6 fusion proteins, various segments of BCL6 cDNA maximum transrepressive activity. These results indicate that were first obtained by polymerase chain reaction (PCR) using the BCL-6 protein can function as a sequence-specific tran- pairs of oligonucleotide primers in which BCL6-specific se- scriptional repressor and have implications for the role of quences are linked to aBamHI restriction site and then subcloned BCL6 in normal lymphoid development and lymphomagenesis. into the BamHI restriction site of the Gal4 plasmid (26). To construct Gal4-BCL-6 fusion protein plasmids with internal The BCL6 gene was identified by virtue of its involvement in BCL6 deletions (constructs J, K, and L; see Fig. 5A), BCL6 chromosomal translocations affecting band 3q27 in non- cDNAs were first obtained by PCR usingBCL6 primers linked to Hodgkin lymphoma (1-5). Subsequent studies have demon- a 5' EcoRI site and a 3' SmaI site and then cloned into the Gal4 strated that rearrangements of the BCL6 gene can be found in plasmid to form an intermediate fusion protein construct. A 30-40% of diffuse large cell lymphoma and 6-11% follicular second BCL6 cDNA obtained using primers linked to a 5' SmaaI lymphoma (6-8). These alterations cause the deregulated site and 3' BamHI site was then cloned into the Smal/BamHI expression of the BCL6 gene by a mechanism called promoter sites of the intermediate fusion protein construct to create the that is the of final construct. All the fusion constructs express Gal4 (amino substitution, juxtaposition heterologous promot- acids 1-147) fused to variable domains ofBCL6 via five residues ers, derived from other chromosomes, to the BCL-6 coding the of the domain (9). Recent evidence also indicates that the 5' non- (PEFPG) encoded by multiple cloning site Gal4 of the BCL6 is altered somatic plasmid. Translation termination is provided by three-frame coding region gene by point termination codons in the Gal4 plasmid. All PCR-derived se- mutations that are found, independent of rearrangements, in quences as well as subcloning junctions were verified by DNA "70% diffuse large cell lymphoma and 45% follicular lym- sequencing. B6BS-TK LUC was constructed by insertion of a phoma (10). Thus, most cases of diffuse large cell lymphoma single copy of the BCL6 binding site (B6BS: GAAAATTCCTA- and a significant fraction of follicular lymphoma carry struc- GAAAGCATA; B.H.Y. etal., unpublished data) into theBamHI tural alterations of the regulatory region of the BCL6 gene, site of TK-LUC. Luciferase reporter constructs (G5-TK-LUC suggesting that deregulated BCL6 expression may be impor- and TK-LUC) were obtained from K. Calame (Columbia Uni- tant for lymphomagenesis (11). versity). TK-CAT reporter constructs [G5(+1000)-TK-CAT and The BCL-6 protein is a 95-kDa nuclear phosphoprotein G5(-750)-TK-CAT]and SV-CAT reporter constructs [G5-SV- detectable at low abundance in multiple tissues and expressed CAT, G5-SV'-CAT, and G5(110)-SV-CAT] were obtained from at high levels exclusively in mature B cells (12-14). Within the L. Lania (27) and F. Rauscher (28), respectively (TK, thymidine B-cell lineage, the expression of the BCL-6 protein is specif- kinase; CAT, chloramphenicol acetyltransferase; LUC, lucif- ically regulated during differentiation since it is only detectable erase; SV, simian virus). in B cells within germinal centers (GCs) but not in pre-GC cells Antisera and Immunoprecipitation. The generation and char- or in differentiated progenies such as plasma cells (12-14). The acterization of polyclonal anti-BCL-6 antisera (N70-6; C73-6) structure of the BCL-6 protein includes six Kruppel-type have been described (12). Polyclonal anti-yeast Gal4 antiserum C-terminal zinc-finger (ZF) motifs (2, 4, 5) which have been was obtained from Upstate Biotechnology (Lake Placid, NY). shown to recognize specific DNA sequences in vitro (15, 16), and Cell Lines, Transient Transfection, and Reporter Gene Assays. a N-terminal POZ (also called ZIN or BTB; refs. 17-19) motif The Mutu I and Mutu III cell lines were obtained from A. shared byvarious ZF molecules including theDrosophila develop- Abbreviations: ZF, zinc-finger; GC, germinal center; CAT, chloram- The publication costs of this article were defrayed in part by page charge phenicol acetyltransferase; SV, simain virus; LUC, luciferase, EMSA, payment. This article must therefore be hereby marked "advertisement" in electrophoretic mobility shift assay; B6BS, BCL-6 binding site. accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 6947 Downloaded by guest on September 29, 2021 6948 Biochemistry: Chang et al. Proc. Natl. Acad. Sci. USA 93 (1996) Rickinson (29). Transfections of Mutu III cells (1.5 x 107) were sera against the N-terminus (Fig. 1) or the C-terminus (not performed by electroporation using Gene Pulser (Bio-Rad) shown) of BCL6 showed that this complex contained BCL-6 apparatus. Transfections in NTera-2 cells (from U. Siebenlist, and that the full-length protein was present in the complex. National Institutes of Health), 293T cells (from J. Krolewski, Analogous EMSA performed on nuclear extracts from a panel Columbia University), and NIH 3T3 cells were performed by of B cells representative of discrete stages of B-cell differen- calcium-phosphate precipitation procedure (30). Assays for LUC tiation indicated complete concordance between formation of and CAT activities were performed as described (30). BCL-6 containing protein-DNA complexes and protein ex- Electrophoretic Mobility Shift Assay (EMSA). Preparation pression (data not shown). As a further control for specificity, of nuclear extracts, EMSA, and antibody-mediated supershift we compared the B6BS-binding complexes in cells (NTera-2; assays were performed as described (30) except that the EMSA lacking BCL6 expression) transfected with a BCL6 expressing reaction buffer was 20 mM Hepes (pH 7.5), 50 mM KCl, 5 mM vector or with a control vector (30). The result (Fig. 1) shows MgCl2, 10 pkM ZnCl2, 4% Glycerol, and 100 ,tg of BSA per ml. that a DNA-binding complex, specific for B6BS (see cold cognate competition) and containing BCL-6 (see supershift RESULTS analysis), is also detectable in BCL6 transfected cells, but not Specific DNA-Binding by BCL-6 in B-Cell Nuclear Extracts. in control-transfected cells. The complex detected in NTera- Using the cyclic amplification and selection of targets (CAST) 2-transfected cells appeared to be similar in migration and method (31), we have identified a 20-bp DNA B6BS (GAA- abundance to the one detected in B-cell lines, suggesting either AATTCCTAGAAAGCATA) capable of binding in vitro that the BCL-6 protein is the only component of the complex translated or bacterially produced BCL6 or that, if a cofactor is present, it is expressed in cell types that (B.H.Y. et al., un- do not express BCL6. Taken together, these results demon- published data). Compared with the sites previously reported strate that (15, 16), B6BS contains additional flanking residues that the B6BS sequence can serve as a DNA target site increase for in vivo synthesized full-length BCL-6 protein. affinity and contains unequivocal residues at various BCL6 Represses Transcription from Promoter Linked to Its previously undefined positions. To determine whether B6BS DNA Target Site. To investigate the role of BCL6 in the could serve as a target for naturally synthesized BCL-6 in regulation of transcription, we analyzed the ability of a BCL6 nuclear extracts, we used it as a probe in an EMSA by using expression vector to affect the transcription of reporter genes nuclear extracts prepared from B-cell lines expressing (Mutu linked to a promoter and the B6BS site in transient transfec- I) or lacking (Mutu III) BCL6 RNA and protein (12). Fig. 1 tion assays in Mutu III cells (Fig. 24). Cotransfection of the shows that a DNA-binding protein complex was detectable BCL6 expression vector with a reporter gene (LUC) driven by only in nuclear extracts from cells expressing BCL6 (Mutu I), the constitutively active TK promoter linked downstream to a but not in control cells (Mutu III). Cold cognate DNA single-copy B6BS led to a strong (90% using 1 pmol effector competition showed that the detected complex bound B6BS plasmid) and dose-dependent repression (Fig. 2B). The tran- specifically; antibody-mediated supershift analysis using anti- srepression activity was dependent upon the presence of the

50 I.0W Pfw 0h A

ii 11 Antiserum - - N70-6 Pre - N70-6 Pre - - N70-6 Pre - N70-6 Pre Cold Oligomer - - + +

BCL-6 ...... * BCL-6

~~~~~~. +......

.: ....

FIG. 1. Identification of DNA-binding complexes containing BCL-6 in nuclear extracts from B-cell lines. Nuclear extracts from Mutu I (subclone of lymphoma line, Mutu BL, positive forBCL6 RNA and protein), Mutu III (a subclone of Mutu BL, negative forBCL6 RNA and protein), NTera-2 transfected with a BCL6 expression vector (NTera-2/BCL-6), or a control vector (NTera-2/control) were assayed for binding to the B6BS probe by EMSA. Arrows point to BCL-6-containing complexes. Cold cognate oligomer competition was performed using a 10-fold excess of the B6BS probe. Supershift analysis was performed by using an anti-BCL-6 (N70-6) antiserum recognizing the N-terminus of BCL6 or a pre-immune (Pre) serum as a control for specificity. Downloaded by guest on September 29, 2021 Biochemistry: Chang et al. Proc. Natl. Acad. Sci. USA 93 (1996) 6949 BCL6 DNA-binding site since a reporter plasmid (TK-LUC) The N-Terminal Half of the BCL-6 Protein Is Capable of lacking B6BS was only marginally affected by BCL6. This Autonomous Transrepression Activity. To conclusively deter- activity is most likely due to nonspecific "squelching" since it mine whether the N-terminal region of BCL6 contains a tran- is detectable independently of the presence of the B6BS site on srepression domain and whether this domain was capable of the reporter gene (not shown). The transrepression activity of acting autonomously, we tested whether the BCL6 transrepres- BCL6 was also dependent upon the presence of an intact sion activity was transferable to an heterologous DNA-binding N-terminal half of the protein since a vector expressing only domain such as the one from the yeast Gal4 transcription factor. the BCL6 ZF domain retained the ability to bind DNA We constructed a vector [Gal4-BCL-6(2-517)] expressing a fu- specifically (not shown), but lost the ability to repress tran- sion protein in which the non-ZF region (amino acids 2-517) of scription. Analogous results were obtained in additional non-B BCL6 was linked to the Gal4 DNA binding domain (amino acids cell lines that do not express BCL6 (293T cells), while trans- 1-147) and cotransfected it (293T cells) with a reporter gene fection assays performed in B-cell lines expressing high levels construct (G5-TK-LUC) driven by the TK promoter region of endogenous BCL6 led to only minimal levels of repression linked to Gal4 binding sites (G5) (Fig. 3A). The results (Fig. 3) (data not shown). These results indicate that BCL6 is a indicate that the Gal4-BCL-6[2-517] fusion protein is capable of site-specific transcriptional repressor and suggest that its tran- strong (90% repression obtained with 0.1 pmol Gal4-BCL-6 srepressive activity is mediated by its N-terminal region. vector) and dose-dependent repression of reporter gene expression. The observed activity is specific since it is dependent upon A the presence of the BCL6 domain in the effector vector (see modest activity of Gal4 vector) and the Gal4 DNA binding site in Reporters the reporter vector (see modest repression of the TK-LUC B6BS-TK-LUC WFr;-F-"LUC reporter gene). Analogous experiments performed using reporter genes driven by minimal promoters (E1B, SV40) lead to TK-LUC F --LUGc some degree of repression (not shown), the specificity of which, however, was difficult to establish due to the low basal activity of these promoters. Taken together, these results indicate that the N-terminal domain of BCL6 (amino acids 2-517) contains a Effectors potent autonomous transrepression domain. 419 517 BCL-6 d[1-418] HA ZF A 121 517 Reporters A.d 7 / // // / ///// POZ ZF G5-TK-LUC FTKX LUC B E BCL-6 TK-LUC C BCL-6 d[1-418] 100 -

.5 25 - 75 Effectors GaI4-

_-( 50 - Gal4 121 517 J Gal4-BCL-6 WJ- 5) (2-517) _ .> GaI4 POZ iZ 25- -ii , C B 0 ...

Effector * TK-LUC Gal4-BCL-6 (2-517) Reporter D Gal4- 125 FIG. 2. BCL6 can repress transcription from a promoter linked to the B6BS site. (A) Schematic representation of the reporter and 5) -6 10(- effector vectors used. B6BS, 1 copy of the B6BS 20 bp site; TK, promoter region of the thymidine kinase gene; LUC, coding region of the luciferase gene; Ad, adenovirus major late promoter region; HA, -Q 75 influenza-hemagglutinin epitope tag. (B) Increasing amounts (0, 0.1, 0.3, 1.0 pmol) of the pMT2-BCL-6 or pMT2-BCL-6 d[1-418] effector Ec50 vectors were cotransfected into Mutu III cells with 0.4 pmol of target plasmid (pB6BS-TK-LUC or TK-LUC) by a modified calcium phos- Cr 25- phate precipitation method. The total amount of transfected DNA and vector sequences was kept constant in each experiment by adding pMT2 DNA to reach 30 ,tg. Two micrograms of a plasmid expressing the bacterial gene were also cotransfected in each experiment Effector 03-galactosidase G5TK-LUC TK-LUC to serve as an internal control for transfection efficiency. Forty-eight Reporter I hours after transfection, cells were harvested and luciferase activity was measured in a luminometer. Values are expressed as percent of those FIG. 3. The N-terminal half of BCL6 contains an autonomous obtained using the reporter vector alone after normalization for ,B-ga- transrepression domain. (A) Schematic representation of reporter and lactosidase values. No significant difference was detectable between the effector vectors used. G5, five copies of the Gal4 binding site (32); SV, basal levels of expression of B6BS-TK-LUC and TK-LUC (not shown). SV-40 promoter-enhancer region; other abbreviations are as in Fig. Experiments were performed in triplicate and error bars are shown. The 2A. (B) Increasing amounts (0, 0.001, 0.003, 0.01, 0.03, 0.1, and 0.3 results indicate that BCL6 expression leads to a dose-dependent repres- pmol) of the Gal4-BCL-6(2-517) or Gal4 effector vector were co- sion of the reporter gene activity and that BCL6-mediated repression is transfected with 2 ,g of the G5-TK-LUC or TK-LUC and 2 ,ug dependent upon the presence of the N-terminal halfofthe BCL-6 protein SV-13-galactosidase (see Fig. 2B) into 293T cells. Values are expressed and the B6BS site on the reporter plasmid. and normalized as described in Fig. 2. Downloaded by guest on September 29, 2021 6950 Biochemistry: Chang et al. Proc. Natl. Acad. Sci. USA 93 (1996) The Transrepression Activity of BCL6 is Position-, Orien- 267-395 in F) was insufficient to confer maximal transrepres- tation-, and Distance-Independent. To characterize further sion activity. the transrepression activity of BCL6, we investigated the The observed reduction in reporter gene activity was not the effects of varying the distance, 5'-3' position and orientation result of differences in transfection or expression efficiencies between the DNA binding site and the promoter of the among the various effector constructs since immunoprecipi- reporter gene. Toward this end, the Gal4-BCL-6[2-517] vector tation analysis of transfectedn cells indicated comparable levels was cotransfected into 293T cells with various constructs in of expression for all Gal4-BCL-6 fusion proteins (representa- which the Gal4 binding site was positioned at variable distance tive results in Fig. SB). The observed results were not due to 5' [750 bp: G5(-750)-TK CAT; 110 bp: G5(-110)-TK CAT], or differences in subcellular localization or DNA binding among 3' [G5(+ 1000)SV-CAT], or at opposite orientations (G5- the various proteins since comparable levels of specific DNA SVCAT versus GS-SV'CAT) with respect to the promoters binding were observed in nuclear extracts from transfected (TK or SV40) driving the CAT reporter gene (Fig. 4). The cells by EMSA using as a probe a synthetic oligonucleotide results show that the transrepression activity was maintained corresponding to the Gal4 binding site (representative results on all reporter constructs tested. Thus, BCL6-mediated tran- are shown in Fig. 5C). Thus, these results indicate that two srepression can act on DNA binding sites independently of noncontiguous domains, corresponding to POZ (amino acids their position (3' versus 5'), orientation and distance from the 11-121) and to amino acids 267-395, are necessary for maxi- promoter of the target gene. mal transrepression by BCL6. Mapping of the BCL6 Transrepression Domain. To determine which region within the N-terminal half (amino acids DISCUSSION 2-517) ofBCL6 was responsible for the transcriptional repres- Based on the presence of ZF motifs and a N-terminal POZ sion activity, we constructed various deletion mutants of the domain, previous studies have suggested that the BCL-6 BCL6 portion of the Gal4-BCL-6[2-517] vector (Fig. SA) and protein may function as a DNA-binding transcription factor (2, tested their ability to repress the expression of the G5-TK- 4, 5). The present study demonstrates that nuclear extracts LUC reporter gene in cotransfection experiments in 293T and from B-cell lines contain protein complexes in which BCL-6 in NIH 3T3 cells. In both cell lines, significant loss of activity binds DNA at it specific target sequence and that BCL6 can was associated with mutants lacking a relatively central domain repress transcription from promoters linked to the same DNA of the BCL6 molecule (amino acids 267-395) (see mutant C target site. These results have implications for the function of versus D), part of the POZ domain (F, G, H, and L), or both BCL-6 and other POZ/zinc finger proteins as well as for the (I). In both cell lines, maximal levels of transrepression activity role ofBCL6 in normal and neoplastic lymphoid development. (>95%), comparable to those obtained with Gal4-BCL-6[2- DNA-Binding Complexes Containing BCL-6 in B Cells. Two 517], were obtained only in those mutants that retained these previous studies have identified DNA consensus sequences ca- two domains, while deletion ofvariable portions between them pable of binding purified recombinant BCL-6 in vitro (15, 16). was functionally irrelevant (J and K). Accordingly, the pres- One of this candidate sites was also shown to bind protein ence of only one of the two domains (POZ in E; amino acids complex in nuclear extracts from cells expressing BCL6, but the Reporters Effectors % Conversion Fold Repression

G5TK CAT GaI4 98.4+/-14.1% 13.3 ...... -I.-,.,.....L. -5Gal4-BCL-6 (2-517) 7.4+1-2.1% G5(+1000)TK CAT GaI4 S.,,. 12.2+/-0.4% ~CAT 8.3 Ga4-BCL-6 (2-517) 1.5+1-1.2% G5 (-750)TK CAT 50.9+/-6.5% .eBg 7.4 ...... GaI4Gal4-BCL-6 (2-517) 6.9+/-0.1%

G5SV CAT GaI4 * .~. 69.2+/-13.0% ...... - 22.7 .CAT| S. l *'''t Gal4-BCL-6 (2-517) 3.1+/-0.2%

G5SVI CAT GaI4 98.0+/-0.3% 20.8 i..:...... l...... Gal4-BCL-6GaI4 (2-517) 4.7+/-1.3%

G5 (110)SV CAT Gal4-BCL-6 (2-517) 28.3+/-4.5% 35.7 ....i...... A ...... 0.8+/-0.3%

FIG. 4. Position-, distance-, and orientiation-independent transcriptional repression by Gal4-BCL-6 fusion proteins. 293T cells were transfected with 0.1 pmol Gal4-BCL-6 (2-517), 1 ,ug SV-f3-galactosidase plasmid, and 2 ,ug of various reporter plasmids (CAT and SV-40 promoter region) as described in Figs. 2 and 3. CAT activity was measured 48 hr after transfection by thin layer chromatography. A representative chromatogram showing the results obtained with each vector combination is shown. Values are expressed as percent conversion of chloramphenicol and as fold-repression obtained by Gal4-BCL-6 versus Gal4 vector with each reporter vector. The results indicate that comparable levels of repression are induced by BCL6 on the TK (7.4-13.3 fold) or SV (20.8-35.7) promoters regardless of the position (3' or 5'), orientation and distance of the Gal4 DNA-binding site from the promoter. Downloaded by guest on September 29, 2021 Biochemistry: Chang et al. Proc. Natl. Acad. Sci. USA 93 (1996) 6951 A

121 240 395 517 681 BCL-6 FO>/Z _ FOZ 1fi

Relative Luciferase Activity (%) 50 aaI 0 25 50 75 100 125 Gal4(1-147) .,,__ _= __ _I._

517 3,3 B L __- 517 _3.8 395 4.1 15.2 *3T3 LII 293T 267

121 I...1~ 122 517 F _--- r 4 g.2 74 Z /Z =-H 57 517 rI -X / Z000001t / 7 _- 57 395 H ______-

57 267 ,9 ,2 2//77/ ... 136 176 395

395 1 36 240 44 }-6.7 I>_j_-~~t6z2283 176 395 -I

29 43 9 B 67 2200k C I 0)" ...... Gal4-BCL-6 (2-517) :. .... 'I Co 0g *.... ,.. .. ?o Gal4-BCL-6 (2-395) ,... lF. .. 9~~

...... Transfection Gal4-BCL-6 (2-267) i; i i ~~( Cold Oligomers - - + 11 - II1 - + II - + Gal4-BCL-6 (57-267) I ...... , vi"X... ,...,..: Gal4-BCL-6 (122-517) *,,. ist.. QZ:,*.. i. .' ,,4 't; Gal4-BCL-6 (57-395) ..0_ .0' :-,,,!w-ji,- X: .:: .:: Si8-.tWz$}...! * l _ * |Stimz Gal4-BCL-6 (57-517) ..'i0,t. | , | | i5b.0 =- Gal4-BCL-6 i5 W!J . i i:::::: ...... ii:si::*: Gal4-BCL-6 (2-395 d[137-1 751) s5 1 'trs l_5'.:.i.'t t5ix i51-. >4] _tit..*Sb .r,*.:*.

*:t!::i i5.1111.-iiliCtti.5-ii-ii5iitzi. io Gal4-BCL-6 (2-395 d[137-239j) -,t< *-l r>"< 44*-t GaI4- ...... v.... . t...... 1IIII.$ i. _ NS

FIG. 5. Mapping of the BCL6 transrepression domain to two noncontiguous regions in the N-terminal half of BCL6. (A) Schematic representation of Gal4-BCL-6 deletion mutant vectors (Right; compare with schematic representation of the BCL-6 protein on top) and their activity in repressing transcription from a cotransfected G5-TK LUC reporter vector (assayed as described in Fig. 3) in 293 and 3T3 cells (Left). (B) Immunoprecipitation analysis of Gal4-...:::::..::t's-zzgzzezand Gal4-BCL-6 ^ .. - iSztz' fusion proteins expressed in transfected 293T cells. Anti-Gal4 antiserum was used to immunoprecipitate [35S]methionine-labeled proteins that were resolved in 10% SDS/PAGE. Bands corresponding to Gal4-BCL-6 fusion proteins are indicated by open circles. Proteins with molecular mass smaller than that predicted for the corresponding fusion protein represent partial degradation products. Only representative data are shown since different fusion proteins required different gels for analysis. The results indicate that the transfected constructs express the Gal4-BCL-6 proteins at comparable levels. (C) Gal4-BCL-6 deletion mutants retain the ability to bind DNA. Nuclear extracts of 293T cells transfected with the Gal4-BCL-6(2-395), Gal4-BCL-6(57-517), Gal4 vectors or mock-transfected (Mock) were analyzed by EMSA using as a probe a 27-bp oligomer containing the Gal4 DNA-binding site (CGGAAGACTCTCCTCCG; ref 32). Fifty-fold cold cognate oligomer was used in the assay to control for specificity. NS, nonspecific band. The results indicate that Gal4-BCL-6(2-395) (active in transrepression; see A) and Gal4-BCL-6(57-517) (inactive) display comparable DNA-binding activities.

presence of BCL-6 within in vivo formed DNA-binding com- specific immunodetection of BCL-6 in DNA binding assays, we plexes could not be demonstrated (16). Using a target DNA provide direct evidence for the presence of BCL-6 within natu- sequence optimized for in vitro binding to BCL-6 as well as rally occurring complexes in various B-cell types (Fig. 1). This Downloaded by guest on September 29, 2021 6952 Biochemistry: Chang et al. Proc. Natl. Acad. Sci. USA 93 (1996) result validates the B6BS site as the best target for BCL6 in vitro the same genes, and thus differentiation and death, may be and suggest that this site may represent the target for BCL6 constitutively repressed in lymphomas in which BCL6 expres- binding in vivo. sion is deregulated by promoter substitution or mutation. The Our results also demonstrate that BCL-6 contacts DNA as results reported in this study provide the experimental frame- a full-length protein in in vivo formed protein complexes. This work for the identification of BCL6 target genes. in view of the observation that another finding is relevant We are grateful to K. Calame, L. Lania, F. Rauscher, and A. POZ/zinc finger protein, ZID, cannot bind DNA as a full- Rickinson for generous gifts of various plasmids and cell lines. This length protein and that binding-inhibition is mediated by the work was supported by National Institutes of Health Grants CA-44029 POZ domain (19). Our results exclude the possibility that (to R.D.-F.) and CA-34775 (to R.S.K.C.). B.H.Y. is supported by a BCL-6 can only bind DNA in a processed form lacking the fellowship from the Leukemia Society of America. POZ domain (19), but leave open the possibility that other 1. Ye, B. H., Rao, P. H., Chaganti, R. S. K. & Dalla-Favera, R. (1993) Cancer proteins may be present in the complex and necessary for DNA Res. 53, 2732-2735. binding by BCL-6. The finding of apparently identical com- 2. Ye, B. H., Lista, F., Lo Coco, F., Knowles, D. M., Offit, K., Chaganti, R. S. K. plexes in a variety of cell types either expressing endogenous & Dalla-Favera, R. (1993) Science 262, 747-750. 3. Baron, B. W., Nucifora, G., McCabe, N., Espinosa, R., III, Le Beau, M. M. or exogenous BCL6 (Fig. 1) would suggest that, if a partner & McKeithan, T. W. (1993) Proc. Natl. Acad. Sci. USA 90, 5262-5266. protein is present, it is widely expressed in various tissue types. 4. Kerckaert, J. P., Deweindt, C., Tilly, H., Quief, S., Lecocq, G. & Bastard, Transcriptional Repression by BCL6. Our results indicate C. (1993) Nat. Genet. 5, 66-70. that the BCL-6 protein can function as a strong transcriptional 5. Miki, T., Kawamata, N., Hirosawa, S. & Aoki, N. (1994) Blood 83, 26-32. 6. Lo Coco, F., Ye, B. H., Lista, F., Corradini, P., Offit, K., Knowles, D. M., repressor. Although variable in potency, transrepressor func- Chaganti, R. S. K. & Dalla-Favera, R. (1994) Blood 83, 1757-1759. tions have been proposed for other POZ/ZF proteins includ- 7. Bastard, C., Deweindt, C., Kerckaert, J. P., Lenormand, B., Rossi, A., ing the B isoform of the yFBP (33), and the ZF5 (17) and PLZF Pezzella, F., Fruchart, C., Duval, C., Monconduit, M. & Tilly, H. (1994) proteins (23). Although it remains possible that, similarly to other Blood 83, 2423-2427. 8. Otsuki, K., Yano, T., Clark, H. H., Bastard, C., Kerckaert, J.-P., Jaffe, E. S. ZF proteins, the transcriptional activity of POZ/ZF proteins may & Raffeld, M. (1995) Blood 85, 2877-2884. be dependent on the cellular environment and may include the 9. Ye, B. H., Chaganti, S., Chang, C.-C., Niu, H., Corradini, P., Chaganti, R. S. ability of transactivate, all POZ/ZF proteins studied so far have K. & Dalla-Favera, R. (1995) EMBO J. 14, 6209-6217. displayed a consistent transrepressive activity in a variety of cell 10. Migliazza, A., Martinotti, S., Chen, W., Fusco, C., Ye, B. H., Knowles, D. M., Offit, K., Chaganti, R. S. K. & Dalla-Favera, R. (1995) Proc. Natl. types and on various promoters, suggesting that site-specific Acad. Sci. USA 92, 12520-12524. transrepression may represent the critical function of these family 11. Dalla-Favera. R., Ye, B. H., Lo Coco, F., Chang, C.-C., Cechova, K., of nuclear factors. During the preparation of this manuscript, Zhang, J., Migliazza, A., Mellado, W., Niu, H., Chaganti, S., Chen, W., Rao, analogous results on the transrepressive activity of BCL6 have P. H., Parsa, N. Z., Louie, D. C., Offit, K. & Chaganti, R. S. K. (1994) Cold Spring Harbor Symp. Quant. Biol. 59, 117-123. also been reported by others (34). 12. Cattoretti, G., Chang, C.-C., Cechova, K., Zhang, J., Ye, B. H., Falini, B. The mechanism by which transrepression occurs remains to Louie, D. C., Offit, K., Chaganti, R. S. K. & Dalla-Favera, R. (1995) Blood be elucidated. Our results suggest that two noncontiguous 86, 45-53. domains are necessary for maximal transrepressive activity of 13. Onizuka, T., Moriyama, M., Yamochi, T., Kuroda, T., Kazama, A., Kanazawa, N., Sato, K., Kato, T., Ota, H. & Mori, S. (1995) Blood 86,25-37. BCL6 and a similar conclusion has been recently reached for 14. Flenghi, L., Ye, B. H., Fizzotti, M., Bigerna, B., Cattoretti, G., Venturi, S., the highly homologous PLZF protein (J. Licht, personal Pacini, R., Pileri, S., Lo Coco, F., Pescarmona, E., Pelicci, P.-G., Dalla- communication). The first domain virtually coincides with Favera, R. & Falini, B. (1995) Am. J. Pathol. 145, 405-411. POZ sequences, an evolutionarily conserved motif that has 15. Baron, B. W., Stanger, R. R., Hume, E., Sadhu, A., Mick, R., Kerckaert, J.-P., Deweindt, C., Bastard, C., Nucifora, G., Aeleznik-Le, N. & McKeithan, T. W. been proposed to mediate protein-protein interactions in large (1994) Genes Chromosomes Cancer 13, 221-224. multimeric complexes (35). The second set of sequences (amino 16. Kawamata, N., Miki, Tk., Ohashi, K., Suzuki, K., Fukuda, T., Hirosawa, S. acids 240-395), not previously identified as a distinct functional & Aoki, N. (1994) Biochem. Biophys. Res. Commun. 204, 366-374. domain, is characterized by a large number of charged amino 17. Numoto, M., Niwa, O., Kaplan, J., Wong, K.-K., Merrell, K., Kamiya, K., Yanagihara, K. & Calame, K. (1993) Nucleic Acids Res. 21, 3767-3775. acids (35/155) and proline (21/155) residues as typically found in 18. Zollman, S., Godt, D., Prive, G. G., Couderc, J.-L. & Laski, F. A. (1994) regulatory regions of transcriptional modulators. One simple Proc. NatL. Acad. Sci. USA 91, 10717-10721. hypothesis to explain the synergistic activity of these two domains 19. Bardwell, V. J. & Treisman, R. (1994) Genes Dev. 8, 1664-1677. is that they may determine the conformation ofBCL6 critical for 20. Harrison, S. D. & Travers, A. A. (1990) EMBO J. 9, 207-216. 21. DiBello, P. R., Withers, D. A., Bayer, C. A., Bayer, C. A., Fristrom, J. W. transrepressive function. Alternatively, it is conceivable that the & Guild, G. M. (1991) Genetics 129, 385-397. two domains may contribute to transrepression through distinct 22. Chardin, P., Courtois, G., Mattei, M. -G. & Gisselbrecht, S. (1991) Nucleic mechanisms. The POZ domain may be involved in the formation Acids Res. 19, 1431-1436. of multimeric protein complexes and their targeting to discrete 23. Chen, Z., Brand, N. J., Chen, A., Chen., S., Tong, J.-H., Wang, Z.-Y., Waxman, S. & Zelent, A (1993) EMBO J. 12, 1161-1167. nuclear substructures (35), while the structure of the second 24. Koonin, E. V., Senkevich, T. G. .& Chernos, V. I. (1992) Trends Biochem. domain and the ability of BCL6 to act in a distance- and Sci. 17, 213-214. orientation- independent manner are more reminiscent of silenc- 25. Xue, F. & Cooley, L.(1993) Cell 72, 681-693. ers interfering with the formation of an active transcription 26. Lillie, J. W. & Green, M. R. (1989) Nature (London) 338, 39-44. 27. Pengue, G., Calabr6, V., Cannada Bartoli, P., Pagliuca, A. & Lania, L. complex (36). The identification of molecules interacting with (1994) Nucleic Acids Res. 22, 2908-2914. these two domains should provide insights into their role in 28. Madden, S. L., Cook, D. M. & Rauscher, F. J., III (1993) Oncogene 8, BCL6-mediated transrepression. 1713-1720. Implications for the Role of BCL6 in Normal Lymphoid 29. Gregory, C. D., Rowe, M. & Rickinson, A. B. (1990) J. Gen. Virol. 71, 1481-1495. Development and Lymphomagenesis. The BCL-6 protein is 30. Chang, C.-C., Zhang, J., Lombardi, L., Neri, A. & Dalla-Favera, R. (1995) expressed in both proliferating and nonproliferating B cells Mol. Cell. Biol. 15, 5180-5188. within GCs, but not in pre-GC cells or in differentiated 31. Wright, E. M., Binder, M. & Funk, W. (1991) Mol. Cell. Biol. 11, 4104-4110. progenies such as plasma cells (12). This pattern suggests that 32. Giniger, E., Varnum, S. & Ptashne, M. (1985) Cell 40, 767-774. 33. Liu, Q., Shalaby, F., Puri, M. C., Tang, S. & Breitman, M. L.(1994) Dev. BCL6 expression may be needed for GC development, while Biol. 165, 165-177. its downregulation may be necessary for differentiation of B 34. Deweindt, C., Albagli, O., Bernardin, F., Dhordain, P., Quief, S., Lantoine, cells into plasma cells. Based on these observations, the finding D., Kerckaert, J. P. & Leprince, D. (1995) Cell Growth Differ. 6, 1495-1503. that BCL6 functions as a site-specific transcriptional repressor 35. Albagli, O., Dhordain, C., Deweindt, C., Lecocq, G., Leprince, D. (1995) suggests Cell Growth Diff 6, 1193-1198. that BCL6 may regulate the GC phenotype by silenc- 36. Cowell, I. G. (1994) Trends Biochem. Sci. 19, 38-42. ing the genes determining plasma cell differentiation or those 37. Chen, W., Zollman, S., Couderc, J.-L. & Laski, F. A. (1995) Mol. Cell. Biol. preventing the death of B cells within the GC. This implies that 15, 3424-3429. Downloaded by guest on September 29, 2021